AI agents aren't 'going rogue.' Frontier labs are skipping security basics

Re:AI agents aren't 'going rogue.' Frontier labs are skippin... by novaRom on Wednesday October 14, 2026 @03:44PM (#50093915)

Why is it not mandatory yet to disclose full trainig data used to produce any new model? Without it we cannot evaluate any product's safety.

They may talk about rogue agents, sandbox escapes, but this will not solve the core issue: secret training recipes are a security hazard in themselves. Full transparency is needed.

[ Parent | on HN ]

Re:AI agents aren't 'going rogue.' Frontier labs are skippin... by andy81 on Wednesday October 14, 2026 @04:15PM (#50094459)

The training data is mostly pirated - avoiding that fact for as long as possible is an attempt to legitimise it and avoid liability.

[ Parent | on HN ]

Re:AI agents aren't 'going rogue.' Frontier labs are skippin... by pixl97 on Wednesday October 14, 2026 @06:47PM (#50096599)

You mean the initial data? That's mostly useless. What? The model shouldn't be allowed access to I robot?

You can't make a generalized model without generalized behavior.

[ Parent | on HN ]